Cassandra ReadTimeout and WriteTimeout in production
In production we are regularly hit by cassandra ReadTimeout and WriteTimeout raised server side.
Write timeouts are usually mitigated by the storage retry proxy but it exist cases where max retries were reached and so origin loading can fail.
Read timeouts are also mitigated by storage retry proxy but it also exist cases where max retries is reached and lead to errors, especially with vault cooking.
Maybe we should consider increasing cassandra server timeout values currently set to default ones ?